Personal Profile & Biography

About Me

I am a PhD candidate working on LLM-based agents for Model-Driven Engineering (MDE). My research focuses on building AI agents that can reliably use specialized engineering tools, selecting the right tool, applying it correctly, and validating the result.

I design the infrastructure that connects LLMs to these tools (via the Model Context Protocol), and develop methods to evaluate and benchmark agent performance at scale. A recurring theme in my work is closing the loop between execution, evaluation, and improvement.

Research Interests

• Agentic Tool Use: Designing LLM agents that select, call, and chain external tools correctly.
• Agent Evaluation & Benchmarking: Building rigorous, reproducible methods to measure agent accuracy, scalability, and failure patterns across versions.
• Synthetic Data Generation: Generating diverse, validated datasets from small seed examples to train and evaluate LLM-based agents.
• Human-in-the-Loop AI: Designing interactive AI systems where users validate, correct, and guide the agent's output step by step.
• Self-Improving Agents: Closing the loop between execution traces and agent specialization, so agents improve automatically over time.
• LLMs for Software Engineering: Applying LLM-based agents to structured, formally specified software engineering tasks and workflows.

Contact Information

• Email: zakaria.hachm@imt-atlantique.fr
• Office: Room 246, Building B, IMT Atlantique, Nantes, France
• Affiliations:
  • IMT Atlantique
  • LS2N (UMR CNRS 6004) - Laboratoire des Sciences du Numérique de Nantes

Online Presence

• GitHub: github.com/zakaria68K
• LinkedIn: linkedin.com/in/zakaria-hachm
• Google Scholar: Scholar Profile